New Microsoft tools connect AI agents with proper data
Semantic modeling to unite enterprise data and automated pipeline capabilities to feed applications with relevant information aim to give autonomous tools appropriate context.
Microsoft on Tuesday unveiled Fabric IQ and Foundry IQ, new capabilities aimed at connecting AI agents with relevant data to give them the context they need to understand an enterprise's operations and take informed actions.
Both were introduced during Ignite, Microsoft's annual user conference in San Francisco.
Fabric IQ, now in preview, is an extension of the semantic layer that gives data context in Power BI, Microsoft's primary analytics platform. Just as semantic models in Power BI give data context to be discovered and operationalized appropriately to inform analysis, Fabric IQ unifies data with an enterprise's operational systems, enabling AI agents to find the necessary data to carry out tasks autonomously.
Foundry IQ, also in preview, feeds AI agents relevant data. The feature is built on Azure AI Search and automates retrieval-augmented generation (RAG) pipelines to deliver agents the proper data.
Given that adhering to a semantic model makes an enterprise's data consistent, semantic modeling is gaining momentum with AI models and applications reliant on high volumes of high-quality data for accuracy. In September, a group of vendors -- including Snowflake and Salesforce -- formed a consortium to establish an open source standard for semantic modeling.
Microsoft is not part of that consortium, but its new features nevertheless provides customers with valuable capabilities, according to David Menninger, an analyst at ISG Software Research.
"Microsoft appears to be pushing its own semantic model in these IQ product previews," he said. "The context of these semantic models is critical to successful AI and BI implementations, particularly agentic AI."
Beyond Fabric IQ and Foundry IQ, Microsoft introduced Azure HorizonDB, a new cloud database service designed to help customers develop and modernize crucial business applications, including AI agents. Additionally, Microsoft launched updates for existing databases.
Aiding agents
Three years after OpenAI's November 2022 launch of ChatGPT marked significant improvement in generative AI (GenAI) technology and sparked surging interest in AI development, agents are now the dominant trend in AI.
Unlike previous AI capabilities, such as chatbots that require human prompts before acting, agents are applications that can be trained to reason and have contextual awareness to act autonomously. With agents now the focus of most AI development, Microsoft aims to make AI development, deployment and management a more efficient process, according to Frank Shaw, the tech giant's chief communications officer.
"As we embark on another Ignite, Microsoft is empowering the complete life cycle of AI, creating tools and solutions to drive the next generation of digital transformation for every organization and at every level of the work they do," he said during a virtual media briefing to highlight some key features revealed during Ignite.
Among them are Fabric IQ and Foundry IQ.
"These two products help AI agents … bridge the gap between raw data and real-world business meaning, and find the context to make decisions," Shaw said.
These new features finally give Microsoft's users a clearer, coherent path from data to agent integration. [However], for organizations already invested in the Microsoft stack, these additions reduce friction more than they add radical new functionality.
Donald FarmerFounder and principal, TreeHive Strategy
Fabric IQ is designed to provide a unified view of an entire enterprise. The semantic model defines business entities and the relationships between them in OneLake, Microsoft's unified data lake in the Fabric environment for analytics and data management; OneLake features integrations with external data management platforms to make all an enterprise's data available in one place.
Once modeled -- including proper governance to ensure security and compliance -- the data is then available to AI agents and analysts to inform decisions and actions.
Foundry IQ, which is integrated with Microsoft Purview to provide proper governance, builds on Fabric IQ by retrieving and delivering relevant data from multiple sources to AI agents.
Together, Fabric IQ and Foundry IQ provide Microsoft users with a way to connect AI agents with appropriate data and are therefore significant new tools, according to Donald Farmer, founder and principal of TreeHive Strategy.
However, he noted that while useful for Microsoft customers, they are not groundbreaking features that represent market-leading innovation.
"These new features finally give Microsoft's users a clearer, coherent path from data to agent integration," Farmer said. "[However], for organizations already invested in the Microsoft stack, these additions reduce friction more than they add radical new functionality. Let's say the updates are helpful rather than transformative."
In addition, Fabric IQ and Foundry IQ will be most beneficial for enterprises that use Microsoft tools throughout their data and AI systems rather than organizations that use tools from a variety of vendors, he continued.
"Companies with minimal commitment to Microsoft data platforms may not gain much benefit," Farmer said. "Teams that have already built highly tuned, domain-specific retrieval systems may see Foundry IQ as too general. They might view it as less flexible than their own pipelines."
Beyond the new features aimed at fueling agentic AI tools, Microsoft introduced HorizonDB, a new PostgreSQL database service designed to better enable users to develop business applications, including agents.
PostgreSQL databases have become more important over the past 18 months as agentic AI development has gained momentum. Flexibility and versatility are two of their primary characteristics, making them a popular choice for developers building AI pipelines calling on disparate types of data.
Given their rising value, Databricks, Snowflake and Redpanda have all acquired SQL database specialists to add PostgreSQL capabilities.
In addition, Microsoft launched an updated version of SQL Server and made Azure DocumentDB and Fabric databases -- a combination of Cosmos DB and a SQL database -- generally available.
Collectively, the new and updated databases are important additions for Microsoft users, according to Menninger.
"The database news is significant in its breadth, touching on a variety of enterprise requirements," he said.
Perhaps most significant, however, is making previously introduced products and updates generally available, Menninger continued.
"For many enterprises, general availability is a prerequisite before a new technology or release can be placed into production," he said. "So, creating a generally available alternative DocumentDB service may be of most interest."
Competitive standing
With new capabilities specifically targeting the data agents use to inform their actions and database capabilities that also address AI development, Microsoft is continuing to add capabilities that address current enterprise needs, according to Menninger.
However, despite the latest additions, he noted that Oracle and Google Cloud may still have the more complete offering at this point.
"There are several fronts on the database and AI battlefield such as multiple database models -- relational, document, graph -- agentic AI capabilities and ease of use," Menninger said. "Hyperscalers, AWS, GCP, Microsoft and Oracle are all addressing these issues, and as a result are tightly clustered near the top of [ISG's ratings], although we give a slide edge to Oracle and Google."
Regarding what Microsoft could do next to continue improving AI development capabilities, Menninger suggested that helping to establish a standard semantic model would benefit not only Microsoft and its users but the data management industry as a whole.
"It would be great to have a standard semantic model that could be shared among different providers," he said. "Whether it is Open Semantic Interchange or something else, it would be great to see Microsoft participating in an effort to establish such a standard."
Farmer likewise noted that while Microsoft is providing its customers with useful capabilities, its tools don't represent the vanguard of AI development.
"These are solid, strategically directional enhancements rather than headline breakthroughs," he said.
Eric Avidon is a senior news writer for Informa TechTarget and a journalist with more than 25 years of experience. He covers analytics and data management.
